test(mcp_impl): pin parseBalancesJson float64 truncation bug (failing)

[gfyrag]

Summary

Failing regression tests demonstrating a silent corruption bug in
internal/mcp_impl/handlers.go::parseBalancesJson. CI is intentionally
red on this branch — see the stacked fix PR below.

JSON numeric values reach the MCP evaluate tool as float64 (Go's
default when unmarshalling into any). The handler converts them with
big.NewFloat(amount).Int(new(big.Int)), which:

• Silently truncates fractional parts. 100.9 becomes 100.
• Silently rounds large values. Anything past ±(2^53 - 1) cannot be
  losslessly represented as a float64, so the integer the caller meant
  is gone before this handler ever sees it.

For a financial DSL these are not acceptable behaviors. The handler
should reject any value that isn't representable as an exact integer
in float64 precision.

Stack

PR	What it does	Target
this PR	Adds failing regression tests	main
Fix PR (TBD)	Validates the amount before conversion	this branch

Merging the fix flips the failing tests green.

Test plan

• TestParseBalancesJsonRejectsNonIntegerAmounts fails today (100.9 is silently truncated to 100).
• TestParseBalancesJsonRejectsUnsafelyLargeAmounts fails today (1e18 is silently accepted).
• TestParseBalancesJsonAcceptsExactIntegerAmounts passes (sanity).

🤖 Generated with Claude Code

[coderabbitai]

Walkthrough

Added unit test file internal/mcp_impl/handlers_test.go containing three tests for parseBalancesJson. Tests verify the parser rejects JSON numeric balances when they are fractional or unsafe-large floats, and accepts safe exact integers by returning the correct *big.Int value.

Changes

parseBalancesJson Test Coverage

Layer / File(s)	Summary
Exact-integer validation tests internal/mcp_impl/handlers_test.go	Three unit tests validate that parseBalancesJson rejects fractional amounts, rejects unsafe-large float64 values beyond 2^53 - 1, and accepts safe exact-integer amounts with correct *big.Int conversion.

🎯 1 (Trivial) | ⏱️ ~3 minutes

🐰 Three tests now guard the balance parser,
Fractional floats shall not pass the marker,
Big integers bloom where exactness reigns,
Safe math prevails through careful constraints! ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the main change: adding tests that demonstrate a float64 truncation bug in parseBalancesJson. It accurately reflects the changeset's purpose.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description check	✅ Passed	The pull request description clearly relates to the changeset by explaining the bug being tested, why the tests fail, and the overall intent to add regression tests.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch test/mcp-float-truncation-bug

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}